Article pubs.acs.org/JACS
Application of a High-Throughput Analyzer in Evaluating Solid Adsorbents for Post-Combustion Carbon Capture via Multicomponent Adsorption of CO2, N2, and H2O Jarad A. Mason,† Thomas M. McDonald,† Tae-Hyun Bae,†,§ Jonathan E. Bachman,† Kenji Sumida,†,⊥ Justin J. Dutton,‡ Steven S. Kaye,‡,∥ and Jeffrey R. Long*,† †
Department of Chemistry, University of California, Berkeley and Materials Science Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States ‡ Wildcat Discovery Technologies Inc., San Diego, California 92121, United States S Supporting Information *
ABSTRACT: Despite the large number of metal−organic frameworks that have been studied in the context of postcombustion carbon capture, adsorption equilibria of gas mixtures including CO2, N2, and H2O, which are the three biggest components of the flue gas emanating from a coal- or natural gas-fired power plant, have never been reported. Here, we disclose the design and validation of a high-throughput multicomponent adsorption instrument that can measure equilibrium adsorption isotherms for mixtures of gases at conditions that are representative of an actual flue gas from a power plant. This instrument is used to study 15 different metal−organic frameworks, zeolites, mesoporous silicas, and activated carbons representative of the broad range of solid adsorbents that have received attention for CO2 capture. While the multicomponent results presented in this work provide many interesting fundamental insights, only adsorbents functionalized with alkylamines are shown to have any significant CO2 capacity in the presence of N2 and H2O at equilibrium partial pressures similar to those expected in a carbon capture process. Most significantly, the amine-appended metal organic framework mmen-Mg2(dobpdc) (mmen = N,N′-dimethylethylenediamine, dobpdc 4− = 4,4′-dioxido-3,3′-biphenyldicarboxylate) exhibits a record CO2 capacity of 4.2 ± 0.2 mmol/g (16 wt %) at 0.1 bar and 40 °C in the presence of a high partial pressure of H2O.
■
pressure and 40−80 °C (Table 1).8 The effects of potentially more reactive gases that are present in lower concentrations,
INTRODUCTION In 2012, coal- and natural gas-fired power plants released 11.1 Gt of carbon dioxidenearly 30% of the total global emissions.1,2 While there are more than 68,000 power plants currently in operation, approximately 300 of these plants are directly responsible for an astonishing 10% of the world’s CO2 emissions. Capturing and permanently sequestering this CO2 would have a significant and immediate impact on rising levels of CO2 in the atmosphere.3,4 With little financial incentive to reduce CO2 emissions in most countries, however, existing carbon capture technologies are simply too expensive to be practical at the scales required for large power plants that release upward of 40 tonnes of CO2 per minute.4,5 Since the most expensive component of any carbon capture and sequestration process is the separation of CO2 from the other gases present in the flue gas of a power plant, a large research effort has focused on developing new materials and processes to remove CO2 from flue gas using as little energy as possible.6,7 While the exact composition of a flue gas depends on the design of the power plant and the source of natural gas or coal, a mixture of mostly N2, CO2, and H2O is released at ambient © XXXX American Chemical Society
Table 1. Expected Range of Compositions for Flue Gas From a Coal- or Natural Gas-Fired Power Plant8 CO2 (mbar) N2 (mbar) H2O (mbar)9
coal
natural gas
120−150 750−800 50−140
30−50 740−800 70−100
such as O2, SOx, NOx, and CO, must also be considered, but, at a minimum, materials are needed that can selectively capture a large amount of CO2 in the presence of N2 and H2O. Taking advantage of the Lewis acidity of CO2, Lewis basic aqueous amine solutions have been studied extensively for extracting CO2 from gas mixtures and are currently used to remove CO2 from many natural gas streams around the Received: January 25, 2015
A
DOI: 10.1021/jacs.5b00838 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX
Article
Journal of the American Chemical Society
Figure 1. Multicomponent adsorption measurements were performed for mixtures of CO2, N2, and H2O in all of the adsorbents shown above as well as zeolite 13X (Na50Al50Si59O218) and an activated carbon (AX-21) that are not pictured. For the metal−organic framework structures, gray, red, blue, dark-yellow, orange, purple, pink, yellow, and bright-green spheres represent C, O, N, Cu, Fe, Zn, F, Si, and Cl atoms, respectively; H atoms have been omitted for clarity. Purple tetrahedra represent Zn atoms, and dark-green spheres represent Mg or Ni atoms. For the zeolite structure (upper right), each vertex represents a tetrahedral SiO4 or AlO4 unit, while teal and dark-orange spheres represent typical positions for extraframework Na and Ca cations, respectively.
world.10 Aqueous amine scrubbers can also be used to capture high-purity CO2 from flue gas, but new materials with lower regeneration energy requirements could lead to a significantly lower overall cost for carbon capture. To this end, solid adsorbents, including zeolites, activated carbons, silicas, and metal−organic frameworks, have received significant attention as alternatives to amine solutions, demonstrating high CO2 capacities and high selectivities for CO2 over N2, together with reduced regeneration energy penalties.11 It is now well established that adsorbents must contain strong CO2 binding sites in order to adsorb a significant amount of CO2 at 50−150 mbar and to achieve the high CO2 purities necessary for cost-effective sequestration.12,13 While many different classes of adsorbents have been studied for CO2 capture, the most promising materials have typically featured exposed metal cations, exposed anions, or alkylamines, all of which can have strong interactions with CO2.11 Despite the large number of adsorbents that have been reported in the context of CO2 capture, the majority of studies
have relied exclusively on pure CO2 and N2 isotherms, which has made it challenging to identify the best materials for capturing CO2 from an actual flue gas mixture that has a significant amount of H2O. This is particularly true for metal− organic frameworks and has hindered progress in the field.14 There have been some noteworthy computational and experimental efforts to evaluate the stability and CO2 capture performance of metal−organic frameworks under more realistic conditions,15,16 but to the best of our knowledge, there are no reports of multicomponent equilibrium adsorption isotherms for mixtures that include CO2, N2, and H2O. In any gas separation application, mixed gas equilibrium adsorption data are critical for comparing the performances of different materials, for designing processes, and for validating theoretical models of mixture adsorption.17 While models such as ideal adsorbed solution theory (IAST) can be used to predict adsorption for simple gas mixtures such as CO2 and N2 with reasonable confidence,18,19 their accuracy is not well established for more complex mixtures such as CO2, N2, and H2O.20 More B
DOI: 10.1021/jacs.5b00838 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX
Article
Journal of the American Chemical Society
Figure 2. (a) In an open system multicomponent adsorption measurement, a mixture of gases is flowed over a packed bed of adsorbent. The flow rate and composition of the inlet and outlet gas streams are recorded until the composition of the outlet gas stream matches the inlet. (b) In a closed system multicomponent adsorption measurement, a mixture of gases is dosed to a sample and allowed to equilibrate. The amount of each component adsorbed is determined from the equilibrium gas-phase composition and either the equilibrium pressure in a volumetric measurement or the equilibrium weight in a gravimetric measurement.
While this assumption is never entirely true, it is a reasonable approximation for many processes and relies on accurate multicomponent equilibrium data.27 Here, we report the design and validation of a highthroughput instrument for the accurate measurement of multicomponent equilibrium adsorption at conditions relevant to post-combustion carbon capture. These measurements are used to compare the performance of 15 different metal−organic frameworks, activated carbons, zeolites, and amine-appended silicas that are representative of the wide range of adsorbents that have been studied for this application (Figure 1).
importantly, all models that rely on single-component adsorption isotherms to predict mixed gas adsorption assume that the adsorbent is in the same thermodynamic state in the presence of each gas. For many adsorbents that exhibit structural or chemical changes specific to different gas molecules, this is most certainly not the case, and direct measurement of mixed gas adsorption is the only way to reliably evaluate gas separation performance. In contrast to single-component adsorption measurements, which are now carried out routinely and with high accuracy using commercial instruments, mixed gas adsorption measurements are often time-consuming, requiring carefully designed custom equipment and complex data analysis.21 As a result, there is a significant lack of mixed gas equilibrium adsorption data reported in the literature.17 The limited mixed gas adsorption data available are mostly for two-component mixtures in zeolites, and equilibrium adsorption data for mixtures of more than two components are exceedingly rare, even though many industrial gas separations involve multicomponent mixtures.17a,22 More routinely, dynamic column breakthrough experiments are used to evaluate the separation performance of an adsorbent by flowing a mixture of gases through a packed bed and measuring the composition of the outlet gas stream as a function of time.23 It is important to note that a typical breakthrough experiment does not yield equilibrium data, and the relationship between breakthrough results and equilibrium adsorption isotherms is not always clear.17c For instance, nearly all adsorbents will show at least some capacity for capturing CO2 in a standard breakthrough experiment with a mixture of CO2, N2, and H2O, since the front of the bed will desiccate the incoming gas mixture, leaving just CO2 and N2 as the gas flows through the bed.24,25 This can lead to misleading conclusions about the intrinsic ability of a material to adsorb CO2 at a specific partial pressure of H2O, particularly when experiments are run on a small amount of sample. Indeed, many factors in addition to multicomponent adsorption capacities can influence experimental breakthrough curves, including column size, column shape, gas flow rates, adsorbent packing density, and extra-column effects.26 Because breakthrough experiments mimic the dynamic conditions of a large-scale separation, they can be helpful in developing processes for CO2 capture. However, multicomponent equilibrium experiments are better suited for comparing the properties of different materials under similar conditions, since the amount of each gas adsorbed is determined only by the partial pressure of each gas and the temperature. Moreover, these equilibrium data can be used to model any dynamic process using local equilibrium theory, where an equilibrium is assumed to exist between the gas and adsorbed phases at every cross-section of an adsorbent bed.
■
RESULTS AND DISCUSSION Multicomponent Adsorption. Although multicomponent adsorption experiments are far less common than singlecomponent experiments in the literature, there has still been significant progress toward developing improved methods for mixed gas measurements. Since it is usually not possible to measure directly the composition of the adsorbed phase, the main challenge in any multicomponent experiment is determining the composition, or relative partial pressures, of the gas phase at equilibrium.17,21 The composition of the adsorbed phase, as well as the amounts of each gas adsorbed, can then be calculated as the difference between the amount of each component added to the system and the amount that is still present in the gas phase at equilibrium. A variety of techniques have been developed for this purpose in both open and closed systems, but preforming measurements with high enough accuracy to provide meaningful results is not trivial and likely explains the lack of published multicomponent data. Often, the adsorption capacities determined from a multicomponent measurement have such high errors that it is impossible to compare the properties of different materials.21 If done accurately, however, any open or closed system measurement should generate equivalent multicomponent adsorption data for a given set of equilibrium conditions. Methods for performing multicomponent measurements have been reviewed thoroughly in the literature,17a,b,21,22 but will be briefly summarized here in the context of choosing an appropriate technique for high-throughput multicomponent adsorption measurements of CO2, N2, and H2O mixtures at conditions representative of a power plant flue gas. In a typical open system experiment, a gas mixture is flowed over a packed bed of adsorbent until the composition of the outlet gas stream is the same as that of the inlet, with a mass spectrometer or gas chromatograph used to record the outlet gas composition (Figure 2a).28 Determining the equilibrium amounts adsorbed of each gas requires the accurate measurement of both the inlet and outlet gas flow rates and compositions as well as appropriate corrections for extraC
DOI: 10.1021/jacs.5b00838 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX
Article
Journal of the American Chemical Society
Figure 3. Simplified schematic of the high-throughput instrument used in this work to perform multicomponent adsorption measurements. Twentyeight independent sample channels share a common gas-dosing manifold and are each connected to the residual gas analyzer (rga) dose volume via a multiposition valve. For multicomponent measurements, a mixture of gases is dosed into the secondary volume and rga dose volume, then expanded to the primary volume, which contains the activated adsorbent. The syringe is cycled multiple times, then the equilibrium pressure is recorded. For mixtures of CO2, N2, and H2O, the dew point transmitter is used to record the partial pressure of H2O, and the rga is used to measure the ratio of the partial pressures of CO2 and N2.
the composition of the dosed gas mixture required to achieve a specific gas-phase composition at equilibrium is difficult to predict.21 In spite of this, closed systems are more amenable to automation and to comparing multicomponent adsorption for a large number of samples with high accuracy. More advanced versions of the open and closed multicomponent experiments described here have also been developed, including the zero length column technique,31 total desorption analysis,32 in situ infrared spectroscopy,33 and the isotope exchange technique.34 While each of these techniques has certain advantages and disadvantages, none are particularly well suited for highthroughput multicomponent measurements with mixtures of CO2, N2, and H2O. High-Throughput Multicomponent Adsorption Instrument. In this work, a closed system approach was used to develop a high-throughput adsorption instrument that can measure multicomponent adsorption for up to 28 samples at a time (Figure S13). The instrument, built by Wildcat Discovery Technologies Inc., has 28 independent sample channels that share a common gas-dosing manifold with inputs for up to 8 gases, including H2O (Figure 3). Each sample channel has a calibrated volume (“secondary volume”) and a 1000 torr pressure transducer (MKS Seta Model 730 absolute capacitance manometer; accuracy = 0.25% reading), which are contained inside a heated enclosure that is maintained at 40 °C to minimize temperature fluctuation. Each secondary volume is connected via 1/16″ stainless steel tubing to a sample chamber (“primary volume”). The secondary volumes are each approximately 21 mL, while the primary volumes are each approximately 14 mL. The secondary volumes are also connected via 1/16″ stainless steel tubing to two Valco multiposition valves, with the 28 channels split evenly between the two valves. The multiposition valves allow each of the 28 channels to be independently opened to the shared 14 mL “rga dose volume” (rga = residual gas analyzer), which contains a mass spectrometer (MKS Microvision 2), a dew point transmitter (Vaisala, accuracy = ± 3 °C), and a 170 mL gastight syringe. All gas lines, dosing volumes, and sample volumes can be heated above 40 °C, allowing H2O dosing pressures of >70 mbar. Custom software allows multicomponent measurements to be performed in a fully automated manner with complete control over all test
column effects and ensuring that the column is isothermal at equilibrium.21 Because the equilibrium gas-phase composition is equivalent to the composition of the inlet gas stream, it is easy to compare the adsorption capacities of different materials under identical equilibrium conditions in an open system multicomponent experiment. Correcting for extra-column effects can, however, be extremely complicated. These corrections are critical to the accuracy of the results, particularly for small sample sizes where the dead volume of the system is not negligible.26a Additionally, it can be challenging to measure the outlet flow rate with high accuracy since the calibration of most flow meters is dependent on the composition of the gas that is flowing through them.26b As a result of these issues, open system measurements often require a large amount of sample in order to collect accurate data, and consequently are not very amenable to high-throughput screening.21 Closed system measurements, on the other hand, are typically more accurate, allowing multicomponent experiments to be performed on smaller quantities of sample.21 Still, there are significant experimental challenges to using closed systems to measure multicomponent equilibrium adsorption in a highthroughput manner. In a typical closed system experiment, a mixture of gases is dosed to an evacuated sample from a calibrated dosing volume, and the gas-phase composition is recorded once the sample has reached equilibrium (Figure 2b).29 Since equilibration times can often be on the order of hours or even days, a circulation pump, or other gas-mixing device that does not alter the amount of gas inside the closed system, is typically needed.30 Similar to open system measurements, a mass spectrometer or gas chromatograph can be used to measure the composition of the gas phase after equilibrium is reached, but now care must be taken to ensure that the equilibrium conditions are not altered when the gas is analyzed.21 The total amount adsorbed can be determined using standard volumetric or gravimetric techniques with calibrated volumes and a pressure transducer or with buoyancy corrections and a balance, respectively. Equilibrium amounts adsorbed can then be determined by material balance using the ideal gas law or an appropriate equation of state. Unlike in open system measurements, it is often challenging to measure multicomponent adsorption consistently at a specific set of equilibrium conditions in a closed system since D
DOI: 10.1021/jacs.5b00838 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX
Article
Journal of the American Chemical Society parameters. To the best of our knowledge, this is the first instrument reported to be capable of performing highthroughput multicomponent adsorption measurements at equilibrium. In order to accurately measure the sample mass, activated samples are loaded in tared 4 mL vials inside a glovebox under a N2 atmosphere. The 4 mL vials are then inserted into aluminum sample assemblies that can each hold up to 7 vials and can be fully sealed while inside the glovebox, with a Schrader valve completing the seal above each sample. The sample assemblies are then transferred to the high-throughput adsorption instrument, and the headspace above each sample is fully evacuated. The instrument then actuates each Schrader valve, opening each sample channel to vacuum. Sample temperatures in the range of 25−150 °C are achieved using heating elements under the sample holders, and temperatures throughout the instrument are continuously recorded by eight thermocouples. Pure-component adsorption isotherms up to a maximum pressure of 1.2 bar are measured using a standard volumetric technique. To ensure that all volume calibrations, pressure transducer calibrations, thermocouple readings, and adsorption calculations are accurate and that any leak rates are negligible, background adsorption isotherms were measured for empty sample holders on all 28 channels using He, N2, CO2, and H2O (Figure S14). The magnitude of the background adsorption was always found to be
